JP4802898B2 - Device and method for correcting an image having linear elements - Google Patents

Description

  The present invention relates to a correction device and a correction method for an image having linear elements, and in particular, corrects the density distribution of linear elements for an image formed by arranging a large number of irregular linear elements called lines or hairlines. Regarding technology.

An image formed by arranging a large number of elongated linear elements is used in various fields. For example,
In many cases, the surface decoration of building materials such as wallpaper and flooring, and the surface decoration of furniture are provided with a pattern in which a large number of linear elements called lines and hairlines are arranged. These patterns may be formed on the surface as an ink layer by normal printing, may be formed as a layer with a metallic luster by a foil pressing method, or formed as an uneven structure on the surface by embossing. Sometimes it is done. In particular, a pattern composed of a large number of linear elements formed by embossing can serve a function of imparting directivity to the reflection of light, and thus is useful in expressing so-called “shine”.

For example, the following Patent Document 1 discloses a technique for embossing a large number of grooves made of lines on a building material surface having a wood grain pattern in order to express the movement of light on the surface of a wood. Patent Document 2 discloses a technique for artificially generating a cross-sectional pattern of a wood grain using a computer. Patent Document 3 defines a two-dimensional scalar field of fiber submerged angles using a computer in order to express the fiber orientation of natural wood, and the lines based on this two-dimensional scalar field. A method for generating an image consisting of a pattern is disclosed. Furthermore, Patent Document 4 discloses a method of expressing an arbitrary three-dimensional pattern by a pattern composed of a number of hairline-shaped grooves.
Japanese Patent Laid-Open No. 5-289302 JP-A-9-327969 JP 2000-351263 A JP 2001-138700 A

  As a pattern formed by arranging a large number of linear elements called lines or hairlines, a standard arrangement of standardized linear elements having the same width and length is also used. In many cases, an irregular arrangement of a large number of linear elements each having an irregular shape is used. In particular, in expressing the natural texture of natural materials such as wood and stone, a pattern formed by arranging a large number of irregularly shaped linear elements is preferably used. As a method for artificially generating such a pattern with a computer, a method for determining the shape and position of each linear element using random numbers, or a shape and position of the linear element using a fractal field are used. A technique for adding natural fluctuations is known.

  However, when an image in which irregular linear elements are irregularly arranged is created by a technique using a computer in this way, there arises a problem that the density distribution of the linear elements is not uniform. Of course, the density distribution of individual pipe cross-sections (linear elements) is not uniform even in the wood grain pipe cross-section patterns that appear on natural wood boards, but in the case of images created by a computer-based technique, they were generated artificially. Since it is a pattern, unnaturalness remains in the density distribution.

  Such an unnatural non-uniformity of density distribution in an image having a linear element is not only a problem in design, but a problem in the manufacturing process when using the image for surface decoration of building materials and packaging materials. Also cause. For example, if there is an extremely dense density for each part of the image, the amount of ink at the time of printing varies from part to part, and printing defects tend to occur. In addition, if the image is transferred by the foil pressing method, the foil layer is likely to be partially peeled off. If the unevenness is formed by embossing, the embossing pressure is partially Every situation is different, and a situation where correct embossing cannot be performed easily occurs.

  Therefore, an object of the present invention is to provide a technique for correcting the density distribution of linear elements so as not to cause a sense of incongruity as a whole for an image in which a large number of irregular linear elements are arranged.

(1) A first aspect of the present invention is an image correction apparatus for correcting a density distribution of linear elements for an image formed by arranging a large number of linear elements.
An original image input unit for inputting an original image to be corrected as a binary image composed of an aggregate of black pixels constituting a linear element portion and white pixels constituting a background portion;
A correction work image storage unit that stores the input original image as an initial image;
A shift image storage unit that stores the input original image as an initial image;
For the image in the correction work image storage unit, a distance calculation unit for obtaining a distance to the closest black pixel for each pixel;
A distance image storage unit for storing a distance image having a value indicating a distance calculated by the distance calculation unit as a pixel value;
A parameter setting unit for setting the predetermined shift amount Δ and the gap width W as parameters;
A shift processing unit that shifts and updates the image in the shift image storage unit by a shift amount Δ in a predetermined direction;
Among the black pixels of the image in the shift image storage unit, the black pixel in which the pixel value of the pixel at the corresponding position of the image in the distance image storage unit exceeds the gap width W is determined as the corresponding position of the image in the correction work image storage unit. A copy processing unit for copying to
Is provided.

(2) According to a second aspect of the present invention, in the image correction apparatus having the linear element according to the first aspect described above,
The copy processing unit performs processing for obtaining the pixel value of the pixel at the corresponding position of the image in the distance image storage unit as each reference pixel value for each black pixel of the image in the shift image storage unit. When a black pixel exceeding the gap width W is found, a process of replacing the white pixel of the image in the correction work image storage unit at the position corresponding to the black pixel with the black pixel is performed.

(3) According to a third aspect of the present invention, in the image correction apparatus having the linear element according to the first aspect described above,
For each white pixel of the image in the correction work image storage unit, the copy processing unit determines that “the pixel of the image in the shift image storage unit at the corresponding position is a black pixel” and “the pixel in the distance image storage unit at the corresponding position”. It is determined whether or not the condition that the pixel value of the pixel of the image exceeds the gap width W is satisfied, and when the condition is satisfied, a process of replacing the white pixel with a black pixel is performed. Is.

(4) According to a fourth aspect of the present invention, there is provided an image correction apparatus for correcting a density distribution of linear elements with respect to an image formed by arranging a large number of linear elements.
An original image input unit for inputting an original image to be corrected as a binary image composed of an aggregate of black pixels constituting a linear element portion and white pixels constituting a background portion;
A correction work image storage unit that stores the input original image as an initial image;
A shift image storage unit that stores the input original image as an initial image;
For the image in the correction work image storage unit, a distance calculation unit for obtaining a distance to the closest black pixel for each pixel;
A parameter setting unit for setting the predetermined shift amount Δ and the gap width W as parameters;
A shift processing unit that shifts and updates the image in the shift image storage unit by a shift amount Δ in a predetermined direction;
Among the black pixels of the image in the shift image storage unit, the black pixel whose distance obtained by the distance calculation unit for the pixel at the corresponding position of the image in the correction work image storage unit exceeds the gap width W is corrected. A copy processing unit for copying to a corresponding position of the image in the image storage unit;
Is provided.

(5) According to a fifth aspect of the present invention, in the image correction apparatus having the linear element according to the first to fourth aspects described above,
The shift processing unit repeatedly performs the shift process only a plurality of times,
The copy processing unit executes a copy process (including an empty copy process where no substantial copy is performed at all) each time the shift process is performed,
The distance calculation unit executes distance calculation processing for the original image and distance calculation processing for the image after the copy processing.

(6) According to a sixth aspect of the present invention, in the image correction apparatus having the linear element according to the fifth aspect described above,
The shift processing unit repeatedly executes the shift processing until the total number of pixels from which the distance exceeding the gap width W is obtained by the distance calculation processing is equal to or less than a predetermined reference value.

(7) According to a seventh aspect of the present invention, in the image correction apparatus having the linear element according to the fifth aspect described above,
The shift processing unit repeatedly executes the shift process until the empty copy process is continuously executed a predetermined number of times.

(8) According to an eighth aspect of the present invention, in the image correction apparatus having the linear element according to the first to seventh aspects described above,
The distance calculation unit obtains the Euclidean distance between the center points of the two pixels or a value obtained by rounding the Euclidean distance by a predetermined effective digit as the calculated value.

(9) According to a ninth aspect of the present invention, in the image correction apparatus having the linear element according to the first to seventh aspects described above,
The distance calculation unit obtains the four-neighbor distance or the eight-neighbor distance between two pixels as a calculation value.

(10) According to a tenth aspect of the present invention, in the image correction apparatus having the linear element according to the first to ninth aspects described above,
The shift processing unit performs a rotation shift in which a pixel protruding from one side of the image by the shift is arranged in a blank portion generated in the other side of the image by the shift.

(11) According to an eleventh aspect of the present invention, in the image correction apparatus having the linear element according to the first to tenth aspects described above,
The parameter setting unit has a function of setting the shift direction as a parameter,
The shift processing unit executes shift processing in the set shift direction.

  (12) According to a twelfth aspect of the present invention, the image correction apparatus having the linear elements according to the first to eleventh aspects described above is realized by incorporating a dedicated program into a computer.

(13) According to a thirteenth aspect of the present invention, in an image correction method for correcting a density distribution of linear elements for an image formed by arranging a large number of linear elements,
The original image to be corrected is input to the computer as a binary image consisting of an aggregate of black pixels forming the linear element portion and white pixels forming the background portion, and the input original image is corrected in the initial state. Original image input stage to be set as an image for use and a shifted image in the initial state;
A parameter setting step for setting the predetermined shift amount Δ and the gap width W as parameters to the computer;
A distance calculating step for obtaining a distance from the nearest black pixel for each pixel of the correction work image in the initial state or the subsequent state;
An image shift stage in which the computer shifts and updates the shifted image in the initial state or the subsequent state by a shift amount Δ in a predetermined direction;
The computer uses, as the corresponding position of the correction work image, the black pixel whose distance obtained in the distance calculation step exceeds the gap width W for the pixel at the corresponding position of the correction work image among the black pixels of the shift image. An image copy stage to copy and update;
A corrected image output stage in which the computer outputs the updated image for correction work as a corrected image;
Is to be executed.

(14) According to a fourteenth aspect of the present invention, in the method for correcting an image having a linear element according to the thirteenth aspect,
The corrected image output stage is executed after the distance calculation stage, the image shift stage, and the image copy stage (including the empty copy stage where no substantial copy is performed) are repeatedly executed a plurality of times. is there.

(15) According to a fifteenth aspect of the present invention, in the correction method for an image having a linear element according to the fourteenth aspect described above,
In the distance calculation stage, the corrected image output stage is executed when the total number of pixels from which the distance exceeding the gap width W is obtained is equal to or less than a predetermined reference value.

(16) According to a sixteenth aspect of the present invention, in the method for correcting an image having a linear element according to the fourteenth aspect described above,
The corrected image output stage is executed when the empty copy stage is continuously executed a predetermined number of times.

(17) According to a seventeenth aspect of the present invention, in the image correction method having a linear element according to the thirteenth to sixteenth aspects described above,
In the distance calculation stage, a Euclidean distance between the center points of two pixels or a value obtained by rounding the Euclidean distance by a predetermined effective digit is obtained as a calculation value.

(18) According to an eighteenth aspect of the present invention, in the image correction method having a linear element according to the thirteenth to sixteenth aspects described above,
In the distance calculation stage, the 4-neighbor distance or the 8-neighbor distance between two pixels is obtained as a calculated value.

(19) According to a seventeenth aspect of the present invention, in the image correction method having a linear element according to the thirteenth to eighteenth aspects described above,
In the image shift stage, a rotation shift is performed in which pixels protruding from one side of the image by the shift are arranged in a blank portion generated in the other side of the image by the shift.

(20) According to a twentieth aspect of the present invention, in the image correction method having a linear element according to the thirteenth to nineteenth aspects described above,
In the parameter setting stage, the gap width W is set to a value substantially equal to the average width value of each linear element.

(21) According to a twenty-first aspect of the present invention, in the image correction method having a linear element according to the thirteenth to twentieth aspects described above,
In the parameter setting stage, a value twice the gap width W is set as the shift amount Δ.

(22) According to a twenty-second aspect of the present invention, in the image correction method having a linear element according to the thirteenth to twenty-first aspects described above,
In the image shift stage, a rough longitudinal direction common to all linear elements is instructed to the computer, and the computer executes a shift process with the direction orthogonal to the longitudinal direction as the shift direction. .

  According to the present invention, for an image in which a large number of irregular linear elements are arranged, all or part of the existing linear elements are copied to an area where the distribution of the linear elements is coarse. It is possible to correct the density distribution without any sense of incongruity.

  Hereinafter, the present invention will be described based on the illustrated embodiments.

<<< §1. Basic concept of the present invention >>
FIG. 1 is a plan view showing an example of an image before correction by the correction method according to the present invention, and FIG. 2 is obtained by applying the correction method according to the present invention to the image shown in FIG. It is a top view which shows an example of the obtained image after correction. The pre-correction image shown in FIG. 1 is an image artificially generated by a computer based on an algorithm using random numbers, and is composed of a large number of irregular linear elements. The pattern on the image before correction is a motif like a vein as a whole, and an algorithm using random numbers is used in the generation process in order to express a natural flow.

  As described above, when an image in which irregular linear elements are irregularly arranged is created using a computer, there is a problem that the density distribution of the linear elements is not uniform. For example, in the example shown in FIG. 1, the void region (white background portion) is conspicuous in some places, and the density distribution of the linear elements is uneven. Such non-uniformity of the density distribution of the linear elements is not desirable in terms of design, and also when used as a surface pattern of building materials or packaging materials by methods such as printing, foil stamping, embossing, etc. As described above, various problems occur.

  An image obtained by applying the correction device and the correction method according to the present invention to the image before correction shown in FIG. 1 is the image after correction shown in FIG. The white void region that was conspicuous in the image of FIG. 1 is eliminated in the image of FIG. 2, and it can be seen that the density distribution of the linear elements is made more uniform. In addition, when the result shown in FIG. 2 is seen, the part to which the correction is made does not appear to be particularly unnatural. Actually, as long as the image of FIG. 2 is viewed, it is impossible to recognize which part is the correction portion, and the overall feeling of incongruity does not occur by applying the correction.

  The feature of the present invention is that the density distribution of linear elements can be corrected so as not to cause a sense of incongruity as a whole for an image formed by arranging a large number of irregular linear elements. The basic principle is that a linear element on the original image to be corrected or a part of the linear element is translated to a gap and copied. According to this basic principle, it is impossible to correct the portion where the linear elements are arranged “densely” in the original image, but new linear elements are added to the portion where the linear elements are arranged “roughly”. As a result, the density can be improved.

  Next, this basic principle will be described in more detail with reference to the plan views of FIGS. Assume that an original image as shown in FIG. 3 is given as a correction target. This original image is a simple image made up of four linear elements L1 to L4, but is actually composed of a set of many pixels. Here, the pixels constituting the portions (hatched portions) of the linear elements L1 to L4 are referred to as “black pixels”, and the pixels constituting the background portion (white portion in the figure) are referred to as “white pixels”. To. “Black pixels” and “white pixels” are distinguished by pixel values. For example, if “black pixel” is represented by a pixel value “1” and “white pixel” is represented by a pixel value “0”, the original image is expressed as an aggregate of a large number of pixels having a 1-bit pixel value. Is done.

  Since there is a gap region SP between the linear elements L1 and L2 of the original image shown in FIG. 3, the distribution density of the linear elements is uneven as a whole. Therefore, for example, if the existing linear element L1 is copied to the void area SP as shown in FIG. 4 and added as a new linear element L1a, the void area SP is eliminated, and the distribution density of the linear elements is increased. Will be made more uniform. In this example, the linear element L1a is an element having exactly the same shape as the original linear element L1, and is a copy of the original linear element L1 translated in the direction indicated by the white arrow in the figure. become.

  However, in practice, it may not be preferable to copy the original linear element L1. For example, consider correcting an original image as shown in FIG. The original image shown in FIG. 5 is obtained by adding the linear element L5 to the original image shown in FIG. 3, but the void region SP is also generated in the left portion of the linear element L5. However, if the existing linear element L1 is copied as it is in order to fill the gap region SP, it is not preferable because it interferes with the linear element L5.

  In such a case, as shown in FIG. 6, only a part of the existing linear element L1 may be copied and added as a new linear element L1b. That is, if only the part of about 2/3 from the left of the original linear element L1 is translated and copied in the direction indicated by the white arrow in the figure, the part of about 1/3 from the right is not copied. Good.

  In this way, if a method of copying an existing linear element or a part of the existing linear element to the void area on the original image to be corrected is adopted, the shape of the copied linear element is the same as that of the existing linear element. Since it conforms to the shape, it is possible to make corrections that do not cause a sense of incongruity as a whole. In addition, by copying the original linear element when copying, the direction of the copied linear element will also conform to the direction of the existing linear element. It is possible to make corrections that do not cause a sense of discomfort overall. In this way, the orientation of the newly added linear element matches the orientation of the existing linear element, and in particular, the orientation is important, such as a line for expressing the “shine” of wood. This is effective in the case of a linear pattern that fulfills the function.

  As described above, in order to perform correction on the computer in accordance with the basic principle that the linear element on the original image to be corrected or a part of the linear element is translated and copied to the gap area, Some kind of algorithm is required. That is, in order for the computer to execute the process of copying the specific part of the specific linear element to the specific part of the gap region, it is necessary to create a program based on a specific algorithm.

  Therefore, in the present invention, this correction processing is executed by the following algorithm. First, for each white pixel of the original image, a distance value with respect to a black pixel (closest black pixel) existing in the spatially closest position is obtained. As a result, a larger distance value is given to the white pixel farther from the existing linear element. Then, a distance image having each distance value as a pixel value is generated. Next, a shift image is generated by shifting the entire original image by a predetermined pitch in a predetermined direction, and among the black pixels on the shifted image, pixels whose pixel values on the corresponding distance image exceed a predetermined value are to be copied. And the density distribution is corrected by copying the linear element composed of the copy target pixels onto the original image. Hereinafter, this algorithm will be described more specifically.

<<< §2. Configuration of correction apparatus according to the present invention >>
Here, a basic configuration of a correction apparatus capable of executing the correction process based on the above-described algorithm will be described with reference to the block diagram of FIG. As illustrated, the correction apparatus includes an original image input unit 10, a shift image storage unit 20, a correction work image storage unit 30, a copy processing unit 40, a shift processing unit 50, a parameter setting unit 60, a distance calculation unit 70, a distance calculation unit 70, The image storage unit 80 is configured to have a function of performing processing for correcting the density distribution of the linear elements on the original image P0 in which a large number of linear elements are arranged.

  However, in practice, this correction device is realized by incorporating a dedicated program into the computer, and each of the above-described components is embodied by a combination of computer hardware and software. Will be. For example, the shift image storage unit 20, the correction work image storage unit 30, and the distance image storage unit 80 are all components having a function of storing image data, but actually, a storage device such as a computer memory or the like. Will be configured. The original image input unit 10 and the parameter setting unit 60 are configured using a computer input device for inputting data and instructions from the outside, and include a copy processing unit 40, a shift processing unit 50, and a distance calculation unit 70. Is realized by a combination of a predetermined arithmetic processing program and an arithmetic processing device.

  The original image input unit 10 is a component that inputs the original image P0 to be corrected as a binary image. As described above, in the image including the linear elements, the pixels constituting the linear element portion are referred to as “black pixels” and the pixels constituting the background portion are referred to as “white pixels”. Therefore, the original image input unit 10 is a component that inputs the original image P0 to be corrected as a binary image composed of an aggregate of black pixels and white pixels.

  Note that the terms “black pixel” and “white pixel” in the present application are simply used as terms meaning “a pixel constituting the linear element portion” and “a pixel constituting the background portion”. This is synonymous with “one attribute pixel” and “second attribute pixel”. Here, instead of the words “first attribute pixel” and “second attribute pixel”, the word “black” or “white” indicating the color is used in the embodiment with reference to the drawings. The “black pixel” and the “white pixel” do not indicate actual colors on the image, for the sake of convenience (in order to facilitate visual understanding).

  The original image P0 input by the original image input unit 10 is stored in the shift image storage unit 20 and the correction work image storage unit 30, respectively. In the embodiment shown here, the original image P0 is data composed of a two-dimensional pixel array of I rows and J columns, and the pixel at the position of the i-th row and j-th column is represented by P0 (i, j). To do. As described above, the pixels constituting the original image P0 are either black pixels or white pixels. That is, the original image P0 is a binary image, and an arbitrary pixel P0 (i, j) has either a pixel value “1” (for black pixels) or a pixel value “0” (for white pixels). Of pixel values.

  The shift image storage unit 20 is a storage area having a function of storing image data composed of a two-dimensional pixel array of I rows and J columns. Here, the image to be stored in the shift image storage unit 20 is referred to as “shift image Sk”, and the pixel at the position of the i-th row and the j-th column of the shift image Sk is defined as Sk (i, j). Here, the subscript “k” in “shift image Sk” and “pixel Sk (i, j)” is an integer of 1, 2, 3,. The shift image Sk is an image obtained by shifting the original original image P0 by a predetermined shift amount by k times in a predetermined direction. If the original image P0 itself is not called a shifted image, the subscript “k” of the shifted image Sk starts from 1 as described above, but the original image P0 itself is shifted in the initial state. If it is grasped as an image (an image obtained by shifting the original image only 0 times), the subscript “k” starts from 0.

  Similarly, the correction work image storage unit 30 is also a storage area having a function of storing image data composed of a two-dimensional pixel array of I rows and J columns. Here, the image to be stored in the correction work image storage unit 30 is referred to as “correction work image Pk”, and is located at the position of the i-th row and j-th column of the correction work image Pk. Let the pixel be represented by Pk (i, j). Here, the subscript “k” is an integer of 0, 1, 2,. Here, for convenience, the original original image P0 itself is also called a correction work image Pk, and the subscript “k” of the correction work image Pk starts from 0.

  Eventually, the original image P0 input from the original image input unit 10 is stored in the shift image storage unit 20 as an initial image for generating a shift image, and in the correction work image storage unit 30. The first correction work image is stored. In other words, the initial image stored in the shift image storage unit 20 and the initial image stored in the correction work image storage unit 30 both match the original image P0.

  The parameter setting unit 60 has a function of setting a predetermined shift amount Δ and a gap width W as parameters. If fixed values are always used as the values of these parameters, the fixed values may be stored in the parameter setting unit 60. However, as will be described later, since this parameter affects the contents of the correction process, it is preferable that the operator can freely set an arbitrary value as a parameter in practice. Specifically, the value of Δ and W may be directly input by an operator from a keyboard or the like, and the numerical value can be increased or decreased by dragging a knob or the like displayed on the display screen with a mouse. An interface may be prepared so that an operator can set a desired numerical value as a parameter by moving a knob or the like to a desired position.

  The shift processing unit 50 executes a process of updating the image stored in the shift image storage unit 20 by shifting the image by a shift amount Δ (value set in the parameter setting unit 60) in a predetermined direction. In the case of the embodiment shown here, since an image composed of a two-dimensional pixel array of I rows and J columns is stored in the shift image storage unit 20, the image is rotated by a shift amount Δ in the downward direction. Is done. For example, when the shift amount Δ is set to 1 pixel, the pixels in the first row move to the second row, the pixels in the second row move to the third row, and the pixels in the third row Moving to the fourth row, the pixels in the I-th row (bottom row) move to the first row (top row). Of course, even if “shifting the image” is actually performed, an operation for moving the pixel values stored in the individual array elements is performed.

  The distance calculation unit 70 calculates a distance from the nearest black pixel for each pixel for the image in the correction work image storage unit 30 and sets a value indicating the distance calculated by the distance calculation unit 70 to the pixel. A distance image Dk having a value is stored in the distance image storage unit 80. This distance image Dk is also image data composed of a two-dimensional pixel array of I rows and J columns, and the pixel at the position of the i-th row and j-th column of the distance image Dk is represented by Dk (i, j). To do. Therefore, the distance image storage unit 80 having a function of storing the distance image Dk is also a storage area having a function of storing image data including a two-dimensional pixel array of I rows and J columns.

  The subscript “k” here corresponds to the subscript “k” of the correction work image Pk, and is also an integer of 0, 1, 2,. For example, the correction work image storage unit 30 stores a correction work image P0 (same as the original image) as an initial image. The calculation performed by the distance calculation unit 70 on the correction work image P0. The image obtained by the above is stored in the distance image storage unit 80 as the first distance image D0. Similarly, the image obtained by the calculation performed by the distance calculation unit 70 on the kth correction work image Pk is stored in the distance image storage unit 80 as the kth distance image Dk. In this manner, the distance image in the distance image storage unit 80 is updated to a new image every time the image in the correction work image storage unit 30 is updated.

  The copy processing unit 40 converts the linear element in the shift image Sk stored in the shift image storage unit 20 or a part thereof onto the correction work image Pk stored in the correction work image storage unit 30. Execute the copy process. Specifically, among the black pixels of the image in the shift image storage unit 20, the pixel value of the pixel at the corresponding position of the image in the distance image storage unit 80 is the gap width W (value set in the parameter setting unit 60). Is determined as a copy target pixel, and a process of copying to a corresponding position of the image in the correction work image storage unit 30 is executed. In other words, it has a function of executing processing for adding the linear element L1a shown in FIG. 4 or the linear element L1b shown in FIG. 6 as a new linear element.

<<< §3. Procedure of correction method according to the present invention >>
Next, the procedure of the correction method according to the present invention will be described with reference to the flowchart of FIG. First, in step S1, an original image input stage is performed. In this stage, the original image to be corrected is input to the computer as a binary image composed of a collection of black pixels constituting the linear element portion and white pixels constituting the background portion. This is a stage of setting as a state correction work image and an initial shift image. For example, in the case of the example shown in FIG. 7, the original image P0 is stored in the shift image storage unit 20 as the shifted image S0 in the initial state, and to the correction work image storage unit 30 as the correction work image P0 in the initial state. Stored.

  In the subsequent step S2, a step of setting the predetermined shift amount Δ and the gap width W as parameters to the computer is performed. In the case of the embodiment shown here, both the shift amount Δ and the gap width W are set as integer values in units of “pixels”. As described above, steps S1 and S2 are processes that should be called preparation steps for causing the computer to execute the correction process. Substantial correction processing procedures are performed in subsequent steps S3 to S5.

  First, in the distance calculation stage of step S3, the computer (distance calculation unit 70) causes the individual pixels of the correction work image Pk (image stored in the correction work image storage unit 30) in the initial state or the subsequent state. Is obtained, and the distance image Dk obtained as a result is stored in the distance image storage unit 80. When Step S3 is first executed following Step S2, the original image P0 is stored in the correction work image storage unit 30 as the correction work image P0 in the initial state. Is obtained by calculation and stored in the distance image storage unit 80.

  At the image shift stage of step S4, the computer (shift processing unit 50) shifts the shift image in the initial state or the subsequent state (image stored in the shift image storage unit 20) by a shift amount Δ (in step S2). A process of shifting and updating by the set parameter) is performed. In the initial state, since the original image P0 is stored in the shift image storage unit 20, the shift image storage unit 20 uses a shift image obtained by rotating the original image P0 by the shift amount Δ. Will be rewritten.

  In the subsequent image copy stage of step S5, the computer (copy processing unit 40) performs correction work image Pk (correction work image storage) for the black pixels of the shift image Sk (image stored in the shift image storage unit 20). The distance (that is, the pixel value in the distance image Dk stored in the distance image storage unit 80) obtained in the distance calculation step for the pixel at the corresponding position of the image stored in the unit 30) Is larger than the gap width W (the parameter set in step S2), the black pixel is copied to the corresponding position of the correction work image Pk and updated.

  For example, let it be assumed that the pixel Sk (i, j) in the i-th row and j-th column in the shift image Sk is a black pixel. In this case, the distance (distance to the closest black pixel) for the pixel Pk (i, j) at the same position in the correction work image Pk is obtained in the distance calculation stage in step S3, and the same position in the distance image Dk. Is recorded as the pixel value of the pixel Dk (i, j). Therefore, the pixel value of the pixel Dk (i, j) is compared with the gap width W, and when the former exceeds the latter, the black pixel Sk (i, j) is placed at the same position in the correction work image Pk. The process of copying to is performed. Actually, a process of changing the pixel value of the pixel Pk (i, j) at the same position in the correction work image Pk to the pixel value “1” indicating the black pixel is performed. If the same processing is performed for all the black pixels in the shift image Sk, the image copy stage in step S5 is completed.

  Through the processing steps of steps S3 to S5, the linear elements are copied from the appropriate positions of the shift image Sk to the void area in the correction work image Pk, and the density distribution of the linear elements is uniform. Will progress. However, in practice, in step S5, no black pixel satisfying the condition (a black pixel to be copied) is found, and there may be a case where no substantial black pixel is copied. In this case, the image copy stage in step S5 is simply an “empty copy stage”. Even if substantial copying is performed in step S5, it is not sufficient in many cases. Therefore, in practice, the distance calculation stage in step S3, the image shift stage in step S4, and the image copy stage in step S5 (including an empty copy stage in which no substantial copy is performed) are repeatedly executed a plurality of times. Is preferred.

  Therefore, in the procedure shown in the flowchart of FIG. 8, a loop is formed through the branch of step S6 and then back to step S3, and the processing of steps S3 to S5 is performed until it is determined “end” in step S6. The steps are executed repeatedly. In step S6, for example, a predetermined condition (whether more practical condition setting will be described later) such as whether or not the number of repetitions has reached a predetermined value set in advance is determined, and the process returns to step S3. Alternatively, it may be determined whether to proceed to step S7.

  In the corrected image output stage of the last step S7, a process of outputting the updated correction work image Pk (the image stored in the correction work image storage unit 30 at that time) from the computer as a corrected image. Is done.

<<< §4. Procedure explanation based on specific example >>>
Here, an example in which the procedure of the correction method shown in FIG. 8 is executed on a specific original image using the correction device shown in FIG. 7 will be described in detail. First, let us consider a case where an original image P0 as shown in FIG. 9 is input in step S1 “original image input stage” of FIG. The original image P0 shown in FIG. 9 is considerably coarser than the actual original image shown in FIG. 1, but for the convenience of explanation, the procedure of the correction process for the original image P0 will be described below. State.

  The original image P0 is an image made up of a two-dimensional pixel array of I rows and J columns, and has three linear elements L11, L12, and L13 as shown in the figure. The small squares shown by hatching in the figure indicate individual black pixels (for example, pixels having a pixel value “1”), and each of the linear elements L11, L12, and L13 is an aggregate of the black pixels. It is constituted by. On the other hand, a region shown as a white background is composed of an aggregate of white pixels (for example, pixels having a pixel value “0”). Here, in order to avoid complication of the figure, the outline of each pixel (small square outline) is displayed only for the black pixel, and the outline of the white pixel is omitted.

  In step S1 “original image input stage” of FIG. 8, the original image P0 shown in FIG. 9 is input from the original image input unit 10 of FIG. 7, and the shift image storage unit 20 and the correction work image storage unit 30 are input. A process of storing as an initial image is performed. Therefore, at this time, the initial shift image S0 stored in the shift image storage unit 20 and the initial correction work image P0 stored in the correction work image storage unit 30 are both illustrated in FIG. That is, the original image P0 shown in FIG.

  In the subsequent step S2 “parameter setting stage”, the shift amount Δ and the gap width W are set as parameters. Here, the following description will be continued assuming that the parameters are set with specific numerical values such as the shift amount Δ = 1 pixel and the gap width W = 2 pixels. Such parameter setting is performed on the parameter setting unit 60 of FIG. 7 as described above.

  Subsequently, the substantial correction work of steps S3 to S5 is started. Steps S3 to S5 are steps that are repeatedly executed as described above. Therefore, the following description will be given step by step for each repetition. As described above, “k” is a variable indicating the number of repetitions.

<First repetition>
First, step S3 “distance calculation stage” is executed. Here, the distance calculation unit 70 obtains the distance to the closest black pixel for each pixel of the correction work image Pk stored in the correction work image storage unit 30, and sets a value indicating the obtained distance to a pixel value. A distance image Dk is created and stored in the distance image storage unit 80. The image currently stored in the correction work image storage unit 30 is the initial correction work image P0, that is, the original image P0 shown in FIG. Therefore, a distance image D0 as shown in FIG. 10 is created and stored in the distance image storage unit 80 for the original image P0 shown in FIG.

  The distance image D0 shown in FIG. 10 is also an image made up of a two-dimensional pixel array of I rows and J columns, and has exactly the same pixel array as the original image P0 shown in FIG. Therefore, an arbitrary pixel D0 (i, j) in the distance image D0 corresponds to a specific pixel P0 (i, j) at the same position in the original image P0 on a one-to-one basis. The pixel value of the pixel D0 (i, j) on the distance image D0 indicates “distance to the nearest black pixel” for the corresponding pixel P0 (i, j) on the original image P0. In addition, on the distance image D0 shown in FIG. 10, the outline (small square) of the black pixel on the original image P0 shown in FIG. 9 is drawn for convenience of explanation. Thereby, the position of the black pixel on the corresponding original image P0 can be easily recognized.

  In FIG. 10, all the numerical values drawn in the outline of the small square are “0” because the nearest black pixel for the black pixel is itself. This is because “” indicates “0”. On the other hand, in FIG. 10, a pixel on which “a numerical value of 1 or more” is drawn is a pixel in which a white pixel is arranged on the corresponding original image P0. The numerical values vary from 1 to 6, but the smaller the numerical value, the more black pixels exist in the vicinity. For example, a pixel in which the numerical value “1” is drawn is a pixel adjacent to a pixel in which the numerical value “0” is drawn, and “distance to the nearest black pixel” is 1. .

  In the case of the embodiment described here, as the “distance to the nearest black pixel”, the Euclidean distance (pure pure) between the center points of two pixels when the pixel arrangement pitch (the vertical and horizontal equal pitches) is a unit distance. Geometric distance). This will be described using the enlarged pixel diagram of FIG.

  FIG. 11 shows a two-dimensional pixel array of 8 rows and 7 columns, in which hatched pixels are black pixels, and white pixels are white pixels. Further, Q1 to Q9 indicate individual pixels and the center points thereof. Here, when the “distance to the nearest black pixel” of each of the pixels Q1 to Q9 is D (Q1) to D (Q9), the value of each distance is a numerical value as shown in the lower part of FIG. .

  For example, D (Q3) = 0. This is because the nearest black pixel for the pixel Q3 is itself, and the Euclidean distance between the center point Q3 and the center point Q3 is 0. It is. Similarly, the distances D (Q4), D (Q6), D (Q7), and D (Q9) for the black pixels Q4, Q6, Q7, and Q9 are all zero. On the other hand, the distance D (Q1) = 2 for the pixel Q1 is that the closest black pixel for the pixel Q1 is Q3, and the Euclidean distance between the center points Q1 and Q3 is 2. Similarly, the distance D (Q2) = 1 for the pixel Q2 is that the closest black pixel for the pixel Q2 is Q3 and the Euclidean distance between the center points Q2 and Q3 is 2. .

  For the pixel Q5, the distance to the black pixel Q6 (the distance between the center points Q5 and Q6) is 2, whereas the distance to the black pixel Q4 (the distance between the center points Q5 and Q4) is √5. Therefore, the closest black pixel is Q6, and the distance D (Q5) = 2. In contrast, for the pixel Q8, the distance to the black pixel Q7 (distance between the center points Q8 and Q7) is 2, whereas the distance to the black pixel Q9 (distance between the center points Q8 and Q9) is 2. Since √2, the closest black pixel is Q9, and the distance D (Q8) = √2.

  As described above, for each pixel, the calculation for obtaining the Euclidean distance with respect to the nearest black pixel is a simple geometric calculation, and various methods are known for performing such calculation on a computer. Then, detailed explanation is omitted. In the embodiment shown here, if a distance value that does not become an integer, such as √2 or √5, is drawn on the distance image Dk as it is, the figure becomes complicated. Therefore, the distance value is rounded off to the nearest decimal point. An example using rounded values will be shown. Of course, in practice, it is preferable to use a value obtained by rounding the original Euclidean distance by a predetermined significant digit as the calculated value of the distance.

  When step S3 “distance calculation stage” is completed in this way, step S4 “image shift stage” is performed. That is, the shift processing unit 50 performs a rotation shift process on the shift image S0 (the same as the original image P0 shown in FIG. 9) currently stored in the shift image storage unit 20. FIG. 12 is a plan view showing a shift image S1 obtained by such a shift process. In the example shown here, since the shift direction is set to the downward direction and the shift amount Δ = 1 pixel, the shift image S1 shown in FIG. 12 is rotated downward by one line from the original image P0 shown in FIG. It has become. The pixel in the I-th row (bottom row) of the shift image S1 is the pixel in the old (I-1) row, and the pixel in the first row (top row) of the shift image S1 is in the old I-row ( (Bottom row) pixels.

  In order to perform such a rotation shift, for example, first, each pixel value of the pixel in the I-th row is saved in a temporary register, and each pixel value of the pixel in the (I-1) -th row is set. , Each pixel value of the pixel in the I-th row is written, each pixel value of the pixel in the (I-2) -th row is written as each pixel value of the pixel in the (I-1) -th row,. ..., each pixel value of the pixel in the first row is written as each pixel value of the pixel in the second row, and finally, the pixel value saved in the register is written in each pixel of the pixel in the first row A process of writing as a value may be performed.

  Finally, step S5 “image copy stage” is executed. This is because the copy processing unit 40 uses the black pixel S1 (i, j) of the image in the shift image storage unit 20 (in this example, the shift image S1 shown in FIG. 12) in the distance image storage unit 80. Black pixel in which the pixel value of the pixel D0 (i, j) at the corresponding position of the image (in this example, the distance image D0 shown in FIG. 10) exceeds the gap width W (in this example, W = 2) This is a process of copying S1 (i, j) to the corresponding position of the image in the correction work image storage unit 30 (in this example, the original image P0 shown in FIG. 9).

  FIG. 13 is a relationship diagram between the shift image S1 and the distance image D0 that are referred to in this copy process. The individual numerical values shown in FIG. 13 are exactly the same as the numerical values indicating the pixel values of the distance image D0 shown in FIG. 10, but the small square outline shown in FIG. 13 is the black pixel of the shift image S1 shown in FIG. It shows the position. In other words, the difference between FIG. 10 and FIG. 13 is that the small square outline shown in FIG. 10 indicates the position of the black pixel of the original image P0, whereas the small square outline shown in FIG. This indicates the position of the black pixel in the shift image S1. As a result, in FIG. 10, the numerical values in the contour are all “0”, whereas in FIG. 13, the numerical values in the contour are a mixture of “0” and “1”.

  In FIG. 13, the copy processing unit 40 searches for pixels whose numerical value in the contour exceeds W = 2 (that is, in this example, the numerical value in the contour is 3 or more), and if there is such a pixel, Using this as a copy target pixel, the image in the correction work image storage unit 30 is copied and updated. In the case of the example shown in FIG. 13, the numerical value in the outline is only 0 or 1, and there are no three or more pixels. Therefore, step S5 “image copy stage” at this point is an empty copy process in which no substantial copy is performed. Here, for convenience, the image for correction work is performed by such an empty copy process. It is assumed that the correction work image P0 stored in the storage unit 30 is updated to the correction work image P1. Of course, the correction work image P1 is the same image as the correction work image P0 (that is, the original image P0), and no data rewriting work is actually performed.

  After all, when the process of step S5 is completed, the shift image S1 shown in FIG. 12 is stored in the shift image storage unit 20, and the original image P0 shown in FIG. 9 is stored in the correction work image storage unit 30. 10 is stored in the distance image storage unit 80, and the distance image D0 shown in FIG. 10 is stored in the distance image storage unit 80.

  Such substantial correction work of steps S3 to S5 is repeatedly executed by the loop through step S6. The subscript “k” of the shift image Sk, the correction work image Pk, and the distance image Dk shown in the block diagram of FIG. 7 corresponds to the number of repetitions. That is, the shift image Sk indicates a shift image obtained after the k-th shift process, and the correction work image Pk is a copy process based on the shift image Sk (empty copy process in which no substantial copy is performed). The distance image Dk is a distance image obtained by the distance calculation for the correction work image Pk.

  Eventually, the processing of steps S3 to S5 is repeatedly executed by a loop through step S6. However, during the k-th repeated execution, the distance calculation processing of step S3 performs the correction work image P (k -1) is a process of performing the distance calculation for the distance image D (k-1) obtained as a result, and storing the distance image D (k-1) in the distance image storage unit 80. This is a process of performing a rotation shift on k-1) and storing the resulting new shift image Sk in the shift image storage unit 20. The copy process in step S5 is a black pixel of the shift image Sk. Among them, the black pixel whose pixel value at the corresponding position of the distance image D (k−1) exceeds the gap width W is determined as the corresponding position of the correction work image P (k−1). Copy, it comes to processing to store the new modified working image Pk obtained as a result of the correction work image storage unit 30.

<Second repetition>
Then, it returns to step S3 from FIG.8 S6, and it continues and demonstrates the process sequence in case the 2nd repetition is performed. First, in step S3 “distance calculation stage”, the distance calculation unit 70 performs distance calculation on the correction work image P1 stored in the correction work image storage unit 30, and uses the obtained distance image D1 as the distance image. A process of storing in the storage unit 80 is performed. However, in the example shown here, the previous copy process was the empty copy process, so the correction work image P1 is exactly the same as the correction work image P0 (that is, the original image P0). Therefore, as a matter of course, the distance image D1 is exactly the same as the distance image D0, and it is not necessary to perform a new distance calculation. Therefore, practically, when the previous copy process was an empty copy process, the substantial calculation process in step S3 “distance calculation stage” can be omitted.

  Subsequently, step S4 “image shift stage” is performed. That is, the shift processing unit 50 performs a rotation shift process on the shift image S1 (FIG. 12) currently stored in the shift image storage unit 20. FIG. 14 is a plan view showing a shift image S2 obtained by such a shift process. As shown in the figure, since the line is shifted downward by one row, the original linear element L13 is divided into three linear elements L13-1, L13-2, and L13-3.

  Then, step S5 “image copy stage” is executed. This is because the copy processing unit 40 uses the distance image D1 (shown in FIG. 10) in the distance image storage unit 80 among the black pixels S2 (i, j) of the shift image S2 (FIG. 14) in the shift image storage unit 20. The black pixel S2 (i, j) in which the pixel value of the pixel D1 (i, j) at the corresponding position of the distance image D0) exceeds the gap width W (W = 2 in this example) is corrected. This is a process of copying to the corresponding position of the correction work image (same as the original image P0 shown in FIG. 9) in the work image storage unit 30.

  FIG. 15 is a diagram showing the relationship between the shift image S2 and the distance image D1 that are referred to in this copy process. The small square outline shown in FIG. 15 indicates the position of the black pixel in the shift image S2 shown in FIG. 14, and the numerical value in this outline is a mixture of “1” and “2”. .

  In FIG. 15, the copy processing unit 40 searches for a pixel whose numerical value in the outline exceeds W = 2 (that is, a numerical value in the outline of 3 or more), and if there is such a pixel, this is copied. As a pixel, the image in the correction work image storage unit 30 is copied and updated. However, in the example shown in FIG. 15 as well, since there are no more than three pixels, an empty copy process is also performed.

  After all, when the second iteration procedure is completed, the shift image S2 shown in FIG. 14 is stored in the shift image storage unit 20, and the original image shown in FIG. 9 is stored in the correction work image storage unit 30. The correction work image P2 that is exactly the same as the image P0 is stored, and the distance image storage unit 80 is in a state in which the distance image D1 shown in FIG. 15 (same as the distance image D0 in FIG. 10) is stored.

<3rd iteration>
Next, returning to step S3 from step S6 in FIG. 8, the processing procedure when the third iteration is executed will be described. First, in step S3 “distance calculation stage”, the distance calculation unit 70 performs a distance calculation on the correction work image P2 stored in the correction work image storage unit 30, and uses the obtained distance image D2 as a distance image. A process of storing in the storage unit 80 is performed. In the example shown here, since the previous copy process was an empty copy process, the correction work image P2 is exactly the same as the correction work image P1 (that is, the original image P0), and the substantial calculation process is omitted. can do.

  Subsequently, step S4 “image shift stage” is performed. That is, the shift processing unit 50 performs a rotation shift process on the shift image S2 (FIG. 14) currently stored in the shift image storage unit 20. FIG. 16 is a plan view showing a shift image S3 obtained by such a shift process. As shown in the figure, the linear element L13 located at the bottom in the original image P0 is located at the top in the shift image S3.

  Then, step S5 “image copy stage” is executed. This is because the copy processing unit 40 uses the distance image D2 (shown in FIG. 10) in the distance image storage unit 80 among the black pixels S3 (i, j) of the shift image S3 (FIG. 16) in the shift image storage unit 20. The black pixel S3 (i, j) in which the pixel value of the pixel D2 (i, j) at the corresponding position of the distance image D0) exceeds the air gap width W (W = 2 in this example) is corrected. This is a process of copying to the corresponding position of the correction work image P2 (same as the original image P0 shown in FIG. 9) in the work image storage unit 30.

  FIG. 17 is a diagram showing the relationship between the shift image S3 and the distance image D2 that are referred to in this copy process. The small square outline shown in FIG. 17 indicates the position of the black pixel in the shift image S3 shown in FIG. 16, and the numerical value in this outline is a mixture of “1”, “2”, and “3”. It has become.

  In FIG. 17, the copy processing unit 40 searches for a pixel whose numerical value in the contour exceeds W = 2 (that is, a numerical value in the contour of 3 or more), and if there is such a pixel, it copies this. As a pixel, the image in the correction work image storage unit 30 is copied and updated. In this example, a pixel whose numerical value in the outline indicates “3” is a copy target pixel. FIG. 18 is a plan view showing only this copy target pixel. Here, for the convenience of explanation, the copy target pixel is indicated by hatching with dots.

  As shown in FIG. 18, these copy target pixels form three elements L21, L22, and L23. These three elements L21, L22, and L23 are partial linear elements to be copied. That is, the partial linear elements L21, L22, and L23 shown in FIG. 18 are elements corresponding to a part of the linear element L11 on the shift image S3 shown in FIG. In the example shown here, the entire linear element L11 was not to be copied because if the entire line was copied, it would be too close to other linear elements (the gap width W could be secured). This is because a part that disappears occurs. This is the same reason that the linear element L1b to be copied is not a whole of the original linear element L1 but a part thereof in FIG.

  Eventually, the copy processing unit 40 causes the partial linear elements L21, L22, and L23 shown in FIG. 18 to correspond to the correction work image P2 in the correction work image storage unit 30 (same as the original image P0 shown in FIG. 9). Processing to copy to the position is performed. FIG. 19 is a plan view of the correction work image P3 obtained by such a copy process. In the figure, linear elements L11, L12, and L13 indicated by hatching with diagonal lines are existing linear elements included in the original correction work image P2 (original image P0), and are hatched with dots. The partial linear elements L21, L22, and L23 shown are newly added linear elements. Of course, in fact, the hatched pixel and the hatched pixel are the same black pixel having the pixel value “1”, and there is no distinction in data.

  Actually, the copy processing as described above by the copy processing unit 40 can be executed by the following two methods. In the first method, for each black pixel S3 (i, j) of the shift image S3 in the shift image storage unit 20, the pixel D2 (i, j) at the corresponding position of the distance image D2 in the distance image storage unit 80 is used. ) Is obtained as each reference pixel value, and when a black pixel S3 (i, j) whose reference pixel value exceeds the gap width W is found, the black pixel S3 (i, j) To replace the white pixel P2 (i, j) of the correction work image P2 in the correction work image storage unit 30 at the position corresponding to the black pixel (rewrite the pixel value “0” to the pixel value “1”). Process). In other words, it can be said to be a method of scanning black pixels on the shifted image side.

  In the second method, for each white pixel P2 (i, j) of the correction work image P2 in the correction work image storage unit 30, “the shift image S3 in the shift image storage unit 20 in the corresponding position” is displayed. Pixel S3 (i, j) is a black pixel (pixel value is “1”) ”and the pixel value of the pixel D2 (i, j) of the distance image D2 in the distance image storage unit 80 at the corresponding position is the gap width It is determined whether or not the condition “exceeds W” is satisfied. If the condition is satisfied, the process of replacing the white pixel P2 (i, j) with a black pixel (pixel value “0” This is a method of performing a process of rewriting to the value “1”). In other words, it is a method of scanning white pixels on the correction work image side.

  After all, when the third iteration procedure is completed, the shift image S3 shown in FIG. 16 is stored in the shift image storage unit 20, and the correction image storage unit 30 shown in FIG. The work image P3 is stored, and the distance image storage unit 80 stores the distance image D2 shown in FIG. 17 (same as the distance image D0 in FIG. 10).

<4th iteration>
Further, the processing procedure when the fourth iteration is executed after returning from step S6 in FIG. 8 to step S3 will be described. First, in step S3 “distance calculation stage”, the distance calculation unit 70 performs a distance calculation on the correction work image P3 stored in the correction work image storage unit 30, and uses the obtained distance image D3 as a distance image. A process of storing in the storage unit 80 is performed. Since the correction work image P3 is an image corrected by adding new linear elements L21 to L23, as shown in FIG. 19, the distance image obtained by the calculation for the correction work image P3. D3 is as shown in FIG. When comparing the distance image D3 shown in FIG. 20 with the distance image D0 shown in FIG. 10, it can be seen that the value of the distance in the vicinity of the part to which the new linear elements L21 to L23 are added is lowered. When the substantial copy processing is executed in this way and the correction work image is substantially corrected, the distance value of the portion where the copy processing has been performed is reduced. By sequentially repeating the correction processing procedure, the pixel value of the distance image can be gradually decreased.

  Subsequently, step S4 “image shift stage” is performed. That is, the shift processing unit 50 performs a rotation shift process on the shift image S3 (FIG. 16) currently stored in the shift image storage unit 20. FIG. 21 is a plan view showing a shift image S4 obtained by such a shift process.

  Then, step S5 “image copy stage” is executed. This is because the copy processing unit 40 uses the distance image D3 (FIG. 20) in the distance image storage unit 80 among the black pixels S4 (i, j) of the shift image S4 (FIG. 21) in the shift image storage unit 20. The black pixel S4 (i, j) in which the pixel value of the pixel D3 (i, j) at the corresponding position exceeds the gap width W (W = 2 in this example) is selected as the correction work image storage unit 30. This is a process of copying to the corresponding position of the correction work image P3 (FIG. 19).

  FIG. 22 is a diagram showing the relationship between the shift image S4 and the distance image D3 that are referred to in this copy process. The small square outline shown in FIG. 22 indicates the position of the black pixel of the shift image S4 shown in FIG. 21, and the numerical value in this outline is a mixture of “0” and “1”. .

  In FIG. 22, the copy processing unit 40 searches for a pixel whose numerical value in the contour exceeds W = 2 (that is, a numerical value in the contour of 3 or more), and if there is such a pixel, copies it. As a pixel, the image in the correction work image storage unit 30 is copied and updated. However, in the case of the example shown in FIG. 22, since there are no more than three pixels, an empty copy process is performed.

  After all, when the fourth iteration procedure is completed, the shift image S4 shown in FIG. 21 is stored in the shift image storage unit 20, and the correction operation shown in FIG. 19 is stored in the correction work image storage unit 30. A correction work image P4 that is exactly the same as the work image P3 is stored, and the distance image D3 shown in FIG. 20 is stored in the distance image storage unit 80.

<5th iteration>
Again, the process returns from step S6 in FIG. 8 to step S3, and the fifth iteration is executed. First, in step S3 “distance calculation stage”, the distance calculation unit 70 performs distance calculation on the correction work image P4 stored in the correction work image storage unit 30, and uses the obtained distance image D4 as a distance image. A process of storing in the storage unit 80 is performed. In this case, since the previous copy process was the blank copy process, the distance image D4 is exactly the same as the distance image D3, and the substantial calculation process in step S3 “distance calculation stage” can be omitted.

  Subsequently, step S4 “image shift stage” is performed. That is, the shift processing unit 50 performs a rotation shift process on the shift image S4 (FIG. 21) currently stored in the shift image storage unit 20. FIG. 23 is a plan view showing a shift image S5 obtained by such a shift process. As shown in the figure, since the line is shifted downward by one line, the original linear element L12 is divided into two linear elements L12-1 and L12-2.

  Then, step S5 “image copy stage” is executed. This is because the copy processing unit 40 uses the distance image D4 (shown in FIG. 20) in the distance image storage unit 80 among the black pixels S5 (i, j) of the shift image S5 (FIG. 23) in the shift image storage unit 20. The black pixel S5 (i, j) in which the pixel value of the pixel D4 (i, j) at the corresponding position of the distance image D3) exceeds the air gap width W (W = 2 in this example) is corrected. This is a process of copying to the corresponding position of the correction work image P4 (same as the correction work image P3 shown in FIG. 19) in the work image storage unit 30.

  FIG. 24 is a diagram showing the relationship between the shift image S5 and the distance image D4 that are referred to in this copy process. The small square outline shown in FIG. 24 indicates the position of the black pixel in the shift image S5 shown in FIG. 23, and the numerical value in this outline is a mixture of “0”, “1”, and “2”. It has become.

  In FIG. 24, the copy processing unit 40 searches for pixels whose numerical value in the contour exceeds W = 2 (that is, those whose numerical value in the contour is 3 or more), and if there is such a pixel, this is copied. As a pixel, the image in the correction work image storage unit 30 is copied and updated. However, in the case of the example shown in FIG. 24, since there are no more than three pixels, an empty copy process is also performed.

  After all, when the fifth iteration procedure is completed, the shift image S5 shown in FIG. 23 is stored in the shift image storage unit 20, and the correction work shown in FIG. 19 is stored in the correction work image storage unit 30. The correction work image P5 that is exactly the same as the work image P3 is stored, and the distance image D4 shown in FIG. 24 (same as the distance image D3 in FIG. 20) is stored in the distance image storage unit 80. Yes.

<6th iteration>
Again, the process returns from step S6 in FIG. 8 to step S3, and the sixth iteration is executed. First, in step S3 “distance calculation stage”, the distance calculation unit 70 performs distance calculation on the correction work image P5 stored in the correction work image storage unit 30, and uses the obtained distance image D5 as a distance image. A process of storing in the storage unit 80 is performed. In the case of the example shown here, since the previous copy process was an empty copy process, the substantial calculation process can be omitted.

  Subsequently, step S4 “image shift stage” is performed. That is, the shift processing unit 50 performs a rotation shift process on the shift image S5 (FIG. 23) currently stored in the shift image storage unit 20. FIG. 25 is a plan view showing a shift image S6 obtained by such a shift process. As shown in the figure, since the line is shifted by one row downward, the original linear element L12 is divided into three linear elements L12-1, L12-2, and L12-3.

  Then, step S5 “image copy stage” is executed. This is because the copy processing unit 40 is in the corresponding position of the distance image D5 in the distance image storage unit 80 among the black pixels S6 (i, j) of the shift image S6 (FIG. 25) in the shift image storage unit 20. A black pixel S6 (i, j) in which the pixel value of the pixel D5 (i, j) exceeds the air gap width W (W = 2 in this example) is used for the correction work in the correction work image storage unit 30. This is a process of copying to the corresponding position of the image P5.

  FIG. 26 is a diagram showing the relationship between the shift image S6 and the distance image D5 that are referred to in this copy process. The small square outline shown in FIG. 25 indicates the position of the black pixel in the shift image S6 shown in FIG. 25, and the numerical values in this outline are [0], “1”, “2”, “3”. ”.

  In FIG. 26, the copy processing unit 40 searches for a pixel having a numerical value in the contour exceeding W = 2 (that is, a numerical value in the contour being 3 or more), and if there is such a pixel, this is copied. As a pixel, the image in the correction work image storage unit 30 is copied and updated. In this example, a pixel whose numerical value in the outline indicates “3” is a copy target pixel. FIG. 27 is a plan view showing only this copy target pixel. Here too, for the convenience of explanation, the copy target pixel is shown hatched with dots.

  As shown in FIG. 27, the current copy target pixel is a partial linear element L31 including only one pixel. Therefore, the copy processing unit 40 moves the partial linear element L31 shown in FIG. 27 to the corresponding position of the correction work image P5 (same as the correction work image P3 shown in FIG. 19) in the correction work image storage unit 30. Processing to copy is performed. FIG. 28 is a plan view of the correction work image P6 obtained by such a copy process. In the figure, linear elements L11, L12, and L13 indicated by hatching with diagonal lines are existing linear elements included in the original image P0, and partial linear elements L21, L22 indicated by hatching with dots. , L23 are linear elements added at the third repetition point described above, and a partial linear element L31 shown by hatching with dots is a linear element added at the current repetition point. Of course, in fact, the hatched pixel and the hatched pixel are the same black pixel having the pixel value “1”, and there is no distinction in data.

  After all, when the sixth iteration procedure is completed, the shift image S6 shown in FIG. 25 is stored in the shift image storage unit 20, and the correction shown in FIG. 28 is stored in the correction work image storage unit 30. The work image P6 is stored, and the distance image storage unit 80 stores the distance image D5 shown in FIG. 26 (same as the distance image D3 in FIG. 20).

<7th iteration>
Subsequently, the process returns from step S6 in FIG. 8 to step S3, and the seventh iteration is executed. First, in step S3 “distance calculation stage”, the distance calculation unit 70 performs distance calculation on the correction work image P6 stored in the correction work image storage unit 30, and uses the obtained distance image D6 as a distance image. A process of storing in the storage unit 80 is performed. As shown in FIG. 28, the correction work image P6 is an image that has been corrected by adding a new linear element L31. Therefore, the distance image D6 obtained by the calculation for the correction work image P6 is As shown in FIG. When the distance image D6 shown in FIG. 29 is compared with the distance image D3 shown in FIG. 20, it can be seen that the distance value in the vicinity of the portion to which the new linear element L31 is added is lowered.

  Here, when the distance image D6 in FIG. 29 is observed, it can be seen that the pixel values of the pixels in the central portion excluding the left and right ends are almost a mixture of “0”, “1”, and “2”. That is, the pixel exceeding the gap width W = 2 is hardly present in the central portion. Thereafter, step S4 “shift stage” and step S5 “image copy stage” are executed to reach step S6. In the embodiment shown here, however, when the seventh iteration is completed, step S6 is completed. From step S7, it is assumed that the loop is repeatedly exited.

  As a result, the corrected image output in step S7 is executed. That is, the correction work image P7 (same as the correction work image P6 shown in FIG. 28) stored in the correction work image storage unit 30 at this time is output as the final corrected image. Become. Eventually, the original image shown in FIG. 9 is corrected to an image as shown in FIG.

  In step S6 in FIG. 8, a determination condition for exiting from the iteration loop returning to step S3 and proceeding to the corrected image output stage in step S7 is, for example, “When the iteration loop has been executed 100 times in total. However, as a more practical condition, the following condition setting can be performed.

  As one condition setting, a condition setting such as “when there is no longer any pixel that can obtain a distance exceeding the gap width W in the distance calculation step S3” can be performed. For example, in the case of the specific example described above, since W = 2 is set, this condition is satisfied when all the pixel values of the distance image Dk obtained in the distance calculation step S3 are 2 or less. become.

  In the case of the specific example described above, as shown in FIG. 9, the original image P <b> 0 does not include linear elements that extend to the left and right ends. Therefore, in the above specific example, even if the loop is repeated infinitely, no new linear element is added near the left and right ends of the image, and “3” or more pixels on the distance image Dk Pixels with values will remain forever. However, if an actual original image (for example, an original image as shown in FIG. 1) is used, such a situation is rare, and the above-described condition setting functions sufficiently in practice.

  However, in order to perform more flexible condition setting, the condition setting such as “when the total number of pixels that can obtain a distance exceeding the gap width W is equal to or less than a predetermined reference value in the distance calculation step S3” is performed. Is preferred. If correction processing based on such a condition setting is performed, the shift processing unit 50 shown in FIG. 7 indicates that the total number of pixels that can obtain a distance exceeding the gap width W in the distance calculation processing in the distance calculation unit 70 is predetermined. It is sufficient to provide a function of “executing the shift process repeatedly until the reference value becomes less than or equal to the reference value”.

  As another condition setting, a condition such as “when an empty copy stage in which no substantial copy is performed at all is continuously executed a predetermined number of times” can be set. In the case of the specific example described above, empty copy was performed in the first and second loops, but substantial copying was performed in the third loop, and further, the fourth and fifth loops. In this case, an empty copy is performed, and a substantial copy is performed in the sixth loop. In this case, the number of continuous executions of the empty copy is two. Therefore, when a condition such as “when the empty copy stage is continuously executed three times or more” is set, the process is executed at least until the eighth loop. If the empty copy is performed in the seventh and eighth loops, and if the empty copy is also performed in the ninth loop, the process exits the loop repeatedly at that time, and the corrected step S7 has been completed. Proceed to the image output stage.

  If correction processing based on such condition setting is performed, the shift processing unit 50 shown in FIG. 7 has a function of “repeatedly execute shift processing until the empty copy processing is continuously executed a predetermined number of times”. Just leave it on. Of course, the image in the correction work image storage unit 30 may be monitored by the operator, and the determination in step S6 may be performed according to the operator's instruction.

<<< §5. Meaning of parameter >>>
In the above §4, the shift amount Δ = 1 and the gap width W = 2 are set as parameters for the original image P0 shown in FIG. 9, and the sixth iterative process (executed at the sixth time) At the stage of performing steps S3 to S5), a correction work image P6 as shown in FIG. 28 was obtained. Here, first, the form of newly added linear elements L21, L22, L23, and L31 (hatched portions by dots) in the image P6 of FIG. 28 will be considered.

  As described above, among these linear elements, L21, L22, and L23 are composed of pixels copied in the third iteration, and L31 is composed of pixels copied in the sixth iteration. It is configured. In both cases, the original linear element L11 in the original image P0 is the copy source.

  That is, L21, L22, and L23 are linear elements obtained by shifting L11 downward by three lines and copying a portion thereof as a copy target. Therefore, when the forms of L21, L22, and L23 are grasped as a whole, it can be seen that the characteristics of the copy source L11 remain. That is, L11 is a linear element having a feature that the left end is curved upward and the right end is curved downward. However, such a feature appears as it is in the entire form of L21, L22, and L23. Yes. L31 is an element obtained by shifting L11 downward by 6 lines and copying a part (only one pixel at the left end) as a copy target. Since there is only one pixel, the feature of the copy source L11 does not appear clearly, but in principle, the feature of L11 is reflected.

  As described above, the basic principle of the present invention is that the linear element on the original image to be corrected or a part of the linear element is translated and copied to the void area, and is based on such a basic principle. If the correction is made, the shape of the copied linear element becomes the same as the shape of the existing linear element, and further, after the parallel movement (shifting downward in the above example) is performed. Since it is a copy, the direction of the copied linear element also conforms to the direction of the existing linear element. For this reason, it is possible to make corrections that do not cause a sense of discomfort overall.

  By the way, in the image P6 shown in FIG. 28, considering the factors that determine the positions and shapes of the newly added linear elements L21, L22, L23, and L31, the shift amount Δ and the gap width set as parameters are set. You can understand that W is greatly involved.

  First, looking at the gaps between the new linear elements L21, L22, L23 and the existing linear element L11, it can be seen that the width of the gap part is two pixels. Similarly, the width of the gap between the new linear element L31 and the existing linear element L12 is also two pixels. This is because the gap width W = 2 is set as a parameter. The reason why the gap width between the new linear elements L21 and L31 is two pixels is that W = 2 is set. Thus, the parameter of the gap width W is literally a parameter that affects the width of the gap between the linear elements. For example, if the gap width W = 1 is set, the linear elements L21, L22, L23 are copied upward by one pixel, and if the gap width W = 3, the linear elements L21, L21, L22 has been copied downward by one pixel.

  On the other hand, the shift amount Δ is a parameter that affects the shape of the new linear element. As described above, the new linear elements L21, L22, and L23 are elements born by copying a part of the existing linear element L11. The process in which these new linear elements L21, L22, and L23 are born can be easily understood with reference to the relationship diagram of FIG. That is, in FIG. 17, a pixel having a pixel value (a value indicating a distance) of 3 or more among pixels having a small square outline (a black pixel on the shift image S3 in FIG. 16) is a copy target pixel. . As a result, as shown in FIG. 18, the linear elements L21, L22, and L23 are extracted as copy targets.

  Here, let us consider a case where the positions of the linear elements L11 to L13 on the shift image S3 shown in FIG. 16 are shifted downward by one pixel. In that case, since the square outlines in FIG. 17 are also shifted down by one pixel, it can be seen that most of the pixel values in the central outline group (the outline group corresponding to L11) are 3 or more. As a result, the linear element extracted as a copy target becomes thicker (has a width corresponding to two pixels).

  Eventually, the gap width W becomes a parameter that governs the width of the gap between each linear element on the corrected image, and the shift amount Δ governs the shape of the newly added linear element on the corrected image. It turns out that it becomes a parameter to do.

  In order to show the function of such parameters more clearly, consider the case where a simple original image P0 as shown in FIG. 30 is given. Linear elements L51 and L52 are arranged on the original image P0 shown in FIG. Each is a linear element composed of a rectangle of 2 pixels in width and 12 pixels in length. The distance image D0 for the original image P0 is as shown in FIG. In FIG. 31, for convenience of explanation, only a part of the distance image D0 (only a part between L51 and L52) is shown.

  FIG. 32 shows a corrected image obtained by performing the correction processing according to the present invention by setting the parameter “shift amount Δ = 1 pixel, gap width W = 2 pixels” for the original image P0 shown in FIG. Is shown. Again, the existing linear elements are hatched with diagonal lines, and the newly added linear elements are hatched with dots. As illustrated, new linear elements L61, L62, and L63 are added. The gap width in the vertical direction of each linear element is approximately 2 pixels (the gap width between L63 and L52 is 4 pixels because there was no room for a new linear element between them. Is). This reflects the setting of the gap width W = 2. In addition, the width of the linear elements L61, L62, and L63 is one pixel, which reflects the setting of the shift amount Δ = 1.

  FIG. 33 shows the result of parameter setting “shift amount Δ = 1 pixel, gap width W = 3 pixels”. As shown in the figure, new linear elements L64 and L65 are added. The gap width in the vertical direction of each linear element is approximately 3 pixels, which reflects the setting of the gap width W = 3. In addition, the width of the linear elements L64 and L65 is one pixel, which reflects the setting of the shift amount Δ = 1.

  FIG. 34 shows the result of parameter setting “shift amount Δ = 1 pixel, gap width W = 4 pixels”. As illustrated, a new linear element L66 is added. The gap width in the vertical direction of the linear elements L51 and L66 is 4 pixels, which reflects the setting of the gap width W = 4. On the other hand, the vertical gap width of the linear elements L66 and L52 is 8 pixels, because there was no room for a new linear element in the gap width W = 4 setting. is there. Further, the width of the linear element L66 is one pixel, which reflects the setting of the shift amount Δ = 1.

  Next, let's look at an example where the shift amount Δ = 2 is set. FIG. 35 shows the result of parameter setting “shift amount Δ = 2 pixels, gap width W = 2 pixels”. As illustrated, new linear elements L71, L72, and L73 are added. The gap width in the vertical direction of each linear element is equal to two pixels, which reflects the setting of the gap width W = 2. Further, the widths of the linear elements L71 and L72 are two pixels, which reflects the setting of the shift amount Δ = 2. The width of the linear element L73 is one pixel because it is close to the existing linear element L52.

  As described above, when the shift amount Δ = 2, the reason why the width of the newly added linear element becomes two pixels can be understood with reference to the distance image D0 in FIG. When the shift amount Δ = 2 is set, a shift image is generated by shifting the existing linear element L51 downward by two lines. In FIG. 31, when the linear element L51 is shifted by two lines, the distance value included in the contour after the shift is “1” or “2”. It cannot be copied. However, if the linear element L51 is shifted by two lines once more (shifted by a total of four lines with respect to the original image), the value of the distance included in the contour after the shift is “3” or “4”. Therefore, all the pixels in the contour are to be copied. For this reason, the pixels for two rows are completely copied, and a linear element having a width of two pixels is added like the linear elements L71 and L72 shown in FIG.

  After all, if the value of the shift amount Δ is set large, the distance value included in the outline after the shift is increased at a time by one shift, and the number of pixels to be copied increases at a stretch, and the relatively wide width. A wide linear element is easily added. In such a point, the shift amount Δ is involved as a parameter that governs the shape of the newly added linear element on the corrected image.

  FIG. 36 shows the result of parameter setting “shift amount Δ = 2 pixels, gap width W = 3 pixels”. As shown in the figure, new linear elements L74 and L75 are added. The gap width in the vertical direction of each linear element is approximately 3 pixels, which reflects the setting of the gap width W = 3. Further, although the shift amount Δ = 2 is set, the width of the linear elements L74 and L75 is one pixel, which depends on the relationship between the shift amount Δ and the gap width W. This is the result. That is, as long as the shift amount Δ = 2 is set, the linear element L51 as a copy source is shifted at an even pitch, but the gap width is set to an odd value of W = 3. Therefore, only one line of the linear element L51 is extracted as a copy target.

  Thus, the shift amount Δ set as a parameter does not necessarily accurately define the width of the newly added linear element, and the width of the actually added linear element is not limited to the line that is the copy source. This is determined by the correlation among the width of the shape element L51, the shift amount Δ, and the gap width W.

  FIG. 37 shows the result of parameter setting “shift amount Δ = 2 pixels, gap width W = 4 pixels”. As illustrated, a new linear element L76 is added. The gap width in the vertical direction of the linear elements L51 and L76 is 4 pixels, and this reflects the setting of the gap width W = 4. On the other hand, the vertical gap width of the linear elements L76 and L52 is 7 pixels, because there was no room for a new linear element in the gap width W = 4 setting. is there. In addition, the width of the linear element L76 is two pixels, which reflects the setting of the shift amount Δ = 2.

  Next, let's look at an example when the shift amount Δ = 3 is set. FIG. 38 shows the result of the parameter setting “shift amount Δ = 3 pixels, gap width W = 2 pixels”. As shown in the figure, new linear elements L81, L82, and L83 are added. The gap width in the vertical direction of each linear element is approximately two pixels, which reflects the setting of the gap width W = 2. Further, the width of the linear elements L81, L82, and L83 is one pixel. This is the set value of the width of the linear element L51 as a copy source, the shift amount Δ = 3, and the gap width W = 2. It was decided by the mutual relationship.

  FIG. 39 shows the result of the parameter setting of “shift amount Δ = 3 pixels, gap width W = 3 pixels”. As illustrated, a new linear element L84 is added. Despite the setting of the gap width W = 3, the vertical gap widths of the actual linear elements L51 and L84 are 4 pixels. This is the linear element L51 that is the copy source. This is because the shift amount Δ = 3 is set even though the width of is only 2 pixels. That is, if a shift amount larger than the width of the linear element L51 as the copy source is set, the copy source does not exist (in the example of FIG. 39, the fourth row, the seventh row, the tenth row, 13th line) occurs, and no pixel is copied to such a line. The reason why the gap width in the vertical direction of the linear elements L84 and L52 is as large as 7 pixels is that the 13th row corresponds to a row in which no copy source exists.

  Finally, let's look at an example where the shift amount Δ = 4 is set. FIG. 40 shows the result of the parameter setting of “shift amount Δ = 4 pixels, gap width W = 2 pixels”. As illustrated, new linear elements L91, L92, and L93 are added. The gap width in the vertical direction of each linear element is approximately two pixels, which reflects the setting of the gap width W = 2. The widths of the linear elements L91, L92, and L93 are determined by the correlation between the width of the linear element L51 that is the copy source and the set values of the shift amount Δ = 4 and the gap width W = 2. .

  FIG. 41 shows the result of parameter setting “shift amount Δ = 4 pixels, gap width W = 3 pixels”. As shown in the figure, new linear elements L94 and L95 are added. The gap width in the vertical direction of each linear element is approximately 3 pixels, which reflects the setting of the gap width W = 3. The widths of the linear elements L94 and L95 are determined by the correlation between the width of the linear element L51 as a copy source and the set values of the shift amount Δ = 4 and the gap width W = 3.

  As can be seen from the above examples, the gap width W set as a parameter is not necessarily accurate, but literally, it is a factor that affects the width of the gap portion between the linear elements. On the other hand, the shift amount Δ set as a parameter is a factor that determines the width of the newly added linear element due to the correlation with the width of the linear element that is the copy source and the gap width W. Become. The operator can obtain a desired correction result by trial and error of arbitrary parameter setting with such prior knowledge.

  According to various trials and errors conducted by the inventors of the present application, a preferable correction result can be obtained by setting a value substantially equal to the average width value of each linear element as the gap width W as the parameter set in the parameter setting stage. It turns out that there is a tendency. In the case of the above-mentioned example, since the width of the linear elements L51 and L52 on the original image is 2 pixels, a preferable result can be obtained by setting W = 2. For example, in the example shown in FIG. 40, it can be seen that new linear elements L91, L92, and L93 are added so as to naturally fuse between the existing linear elements L51 and L52.

  It has also been found that when the shift amount Δ is set to a value twice the gap width W, a preferable correction result tends to be obtained. For example, when the gap width W = 2 is set, it is preferable to set the shift amount Δ = 4. Specifically, the example shown in FIG. 40 is an example in which such a setting is performed. If the various correction results shown in FIGS. 32 to 41 are compared with each other, it can be recognized that the correction result shown in FIG. 40 is the most natural result.

<<< §6. Modified example >>>
Finally, embodiments according to some modifications of the present invention will be described.

(1) Form not using the distance image storage unit 80 In the embodiment shown in the block diagram of FIG. 7, the calculation result by the distance calculation unit 70 is stored as the distance image Dk in the distance image storage unit 80, and the copy processing is performed. The unit 40 refers to the distance image Dk in the distance image storage unit 80 and uses the value of the distance (distance to the closest black pixel) obtained for a specific pixel. However, the distance image storage unit 80 is not necessarily an essential component for carrying out the present invention. If there is no distance image storage unit 80, the calculation result by the distance calculation unit 70 cannot be stored in the format of the distance image Dk. However, in response to an inquiry from the copy processing unit 40, the distance calculation unit 70 Image correction based on the basic principle of the present invention can be carried out by performing processing for calculating the “distance to the nearest black pixel” for a specific pixel each time and reporting the result to the copy processing unit 40.

  That is, it is sufficient that the distance calculation unit 70 has a calculation function for obtaining the distance to the nearest black pixel for each pixel in the image in the correction work image storage unit 30, and the copy processing unit 40 performs the shift. Among the black pixels of the image in the image storage unit 20, the black pixel whose distance obtained by the distance calculation unit 70 for the pixel at the corresponding position of the image in the correction work image storage unit 30 exceeds the gap width W Is copied to the corresponding position of the image in the correction work image storage unit 30.

(2) Order of Steps in FIG. 8 The procedure of the image correction method according to the present invention has been described based on the flowchart of FIG. 8, but the procedure of the flowchart need not necessarily be performed in this order. For example, it is possible to execute the parameter setting step of step S2 before performing the original image input step of step S1, or to perform the image shift step of step S4 before performing the distance calculation step of step S3. Is also possible. In short, the order of the steps in this flowchart can be changed as appropriate without departing from the basic principle of the present invention.

(3) Distance to nearest black pixel As the “distance to nearest black pixel” calculated for each pixel by the distance calculation unit 70, in the embodiments described so far, the Euclidean between the center points of the two pixels is described. A value obtained by rounding the distance or the Euclidean distance by a predetermined significant digit (in the case of the above-described embodiment, a value obtained by rounding off the decimal point) has been used as a calculation value. However, it is not always necessary to use the Euclidean distance as the “distance to the nearest black pixel”. For example, a four-neighbor distance or an eight-neighbor distance of two pixels can be substituted.

  Here, the four-neighbor distance is used as a value indicating the distance between any pixels in a pixel array consisting of a two-dimensional matrix, and is one by one vertically or horizontally on the grid of the two-dimensional matrix. It is a distance indicated as a numerical value indicating how many times the movement can reach the pixel of the other party when moving. For example, in the example shown in FIG. 11, the four-neighbor distance between the pixels Q5 and Q6 is 2, but the four-neighbor distance between the pixels Q5 and Q4 is 3. This is because a total of three movements are required to move from the square of pixel Q5 to the square of pixel Q4 by moving one by one vertically or horizontally. Similarly, the four neighboring distances between the pixels Q8 and Q9 are 2 because they cannot move diagonally.

  On the other hand, the 8-neighbor distance is a value indicating the number of times of movement when further oblique movement is permitted. In the above example, the 8-neighbor distance between the pixels Q5 and Q4 is 2, and the 8-neighbor distance between the pixels Q8 and Q9 is 1.

  Since the correction processing performed in the present invention is for correcting the density distribution of the linear elements by copying all or part of the existing linear elements in the void area on the original image, the distance calculation In the unit 70, a strict distance calculation is unnecessary. Therefore, when the present invention is implemented, there is no great problem even if a convenient distance such as the above-mentioned four-neighbor distance or eight-neighbor distance is used as the “distance to the nearest black pixel”.

(4) Shift processing without rotation In the embodiments so far, the example in which the shift processing unit 50 performs the rotation shift on the original image or the shift image has been described. This rotation shift eventually resulted in pixels that protruded from one side of the image due to the shift (in the case of the previous embodiment, one row of pixels that protruded from the lower end of the image) to the other side of the image due to the shift. This is a shift arranged in a blank portion (in the case of the above-described embodiment, a blank portion corresponding to one line generated at the upper end of the image), which is a preferable shift form in order to effectively use pixels.

  However, the shift processing in the present invention does not necessarily have to be performed by rotation shift. For example, pixels protruding from one side of the image may be discarded as they are. In this case, the blank portion gradually spreads from the other side of the image, but when the total shift amount is small compared to the size of the image, there is no big trouble in implementing the present invention. .

(5) Method of setting parameters for the shift direction In the embodiments described so far, the shift direction is determined to be the downward direction, but it is also possible to arbitrarily set the shift direction as a parameter. In this case, the parameter setting unit 60 may have a function of setting the shift direction as a parameter, and the shift processing unit 50 may execute the shift process in the set shift direction. Shifting in an oblique direction may be performed.

  All of the original images shown in the embodiments so far are images including linear elements extending in the horizontal direction. However, for images including linear elements extending in the vertical direction, the shift direction is set to the right. Or it is more effective to set it to the left. Therefore, in practice, the general longitudinal direction common to all linear elements on the original image input by the operator is visually confirmed, the rough longitudinal direction is instructed to the computer, and set in the parameter setting unit 60 as a parameter. Then, it is preferable that the shift processing unit 50 performs a shift process in which the direction orthogonal to the longitudinal direction is the shift direction. However, in carrying out the present invention, it is not essential that the longitudinal direction of the linear element and the shift direction are orthogonal to each other, and sufficient correction is possible even if they are not orthogonal.

It is a top view which shows an example of the image before correction by the correction method which concerns on this invention. It is a top view which shows an example of the image after correction obtained by applying the correction method based on this invention with respect to the image shown in FIG. It is a top view which shows the simple original image for demonstrating the basic concept of this invention. It is a top view which shows the state which added the correction based on the basic concept of this invention with respect to the original image shown in FIG. It is a top view which shows another simple original image for demonstrating the basic concept of this invention. It is a top view which shows the state which added the correction based on the basic concept of this invention with respect to the original image shown in FIG. 1 is a block diagram illustrating a basic configuration of an image correction apparatus according to an embodiment of the present invention. It is a flowchart which shows the basic procedure of the correction method of the image which concerns on one Embodiment of this invention. It is a top view which shows the original image P0 for demonstrating the correction procedure of the image which concerns on this invention. FIG. 10 is a plan view showing a distance image D0 obtained for the original image P0 shown in FIG. 9 (a small square frame indicates the position of a black pixel in the original image P0). It is an enlarged pixel figure which shows the concept of the distance value of each pixel in this invention. FIG. 10 is a plan view showing a shifted image S1 obtained by rotating the original image P0 shown in FIG. 9 downward by one pixel. It is a top view which shows the relationship between the shift image S1 shown in FIG. 12, and the distance image D0 shown in FIG. 10 (a small square frame shows the position of the black pixel of the shift image S1). FIG. 13 is a plan view showing a shift image S2 obtained by rotating the shift image S1 shown in FIG. 12 downward by one pixel. It is a top view which shows the relationship between the shift image S2 shown in FIG. 14, and the distance image D1 (same as the distance image D0 shown in FIG. 10) (a small square frame shows the position of the black pixel of the shift image S2). FIG. 15 is a plan view showing a shift image S3 obtained by rotating the shift image S2 shown in FIG. 14 downward by one pixel. FIG. 17 is a plan view showing a relationship between a shift image S3 shown in FIG. 16 and a distance image D2 (same as the distance image D0 shown in FIG. 10) (a small square frame indicates the position of a black pixel in the shift image S3). It is a top view which shows the black pixel (partial linear element L21, L22, L23) used as a copy object among the black pixels which comprise the shift image S3 shown in FIG. FIG. 19 is a plan view showing a correction work image P3 obtained by copying the black pixels (partial linear elements L21, L22, L23) shown in FIG. 18 onto the original image shown in FIG. It is a top view which shows the distance image D3 calculated | required about the image P3 for correction work shown in FIG. FIG. 17 is a plan view showing a shift image S4 obtained by rotating the shift image S3 shown in FIG. 16 downward by one pixel. It is a top view which shows the relationship between the shift image S4 shown in FIG. 21, and the distance image D3 shown in FIG. 20 (a small square frame shows the position of the black pixel of the shift image S4). FIG. 22 is a plan view showing a shift image S5 obtained by rotating the shift image S4 shown in FIG. 21 downward by one pixel. It is a top view which shows the relationship between the shift image S5 shown in FIG. 23, and the distance image D4 (same as the distance image D3 shown in FIG. 20) (a small square frame shows the position of the black pixel of the shift image S5). FIG. 24 is a plan view showing a shift image S6 obtained by rotating the shift image S5 shown in FIG. 23 downward by one pixel. FIG. 26 is a plan view showing the relationship between a shift image S6 shown in FIG. 25 and a distance image D5 (same as the distance image D3 shown in FIG. 20) (a small square frame indicates the position of a black pixel in the shift image S6). It is a top view which shows the black pixel (partial linear element L31) used as a copy object among the black pixels which comprise shift image S6 shown in FIG. It is a top view which shows the image P6 for correction work obtained by copying the black pixel (partial line element L31) used as the copy object shown in FIG. 27 on the image P3 for correction work shown in FIG. It is a top view which shows the distance image D6 calculated | required about the image P6 for correction work shown in FIG. It is a top view which shows the original image P0 for demonstrating the meaning of parameters (DELTA) and W in the correction procedure of the image which concerns on this invention. FIG. 31 is a plan view showing a part of a distance image D0 obtained for the original image P0 shown in FIG. 30 (L51 and L52 indicate linear elements on the original image P0 as they are, and numerical portions indicate distance images). FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 1 and W = 2 with respect to the original image P0 shown in FIG. 30. FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 1 and W = 3 for the original image P0 shown in FIG. 30. FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 1 and W = 4 for the original image P0 shown in FIG. 30. FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 2 and W = 2 with respect to the original image P0 shown in FIG. 30. FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 2 and W = 3 for the original image P0 shown in FIG. 30. FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 2 and W = 4 for the original image P0 shown in FIG. 30. FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 3 and W = 2 with respect to the original image P0 shown in FIG. 30. FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 3 and W = 3 for the original image P0 shown in FIG. 30. FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 4 and W = 2 to the original image P0 shown in FIG. 30. FIG. 31 is a plan view showing a corrected image Pn obtained by setting parameters Δ = 4 and W = 3 for the original image P0 shown in FIG. 30.

Explanation of symbols

10: Original image input unit 20: Shift image storage unit 30: Correction work image storage unit 40: Copy processing unit 50: Shift processing unit 60: Parameter setting unit 70: Distance calculation unit 80: Distance image storage unit Dk: Distance image (K = 0, 1, 2, 3, ...)
Dk (i, j): Pixels D (Q1) to D (Q9) constituting the distance image: Distance values L1 to L5 of the respective pixels: Linear elements L1a and L1b: Linear elements L11 to L13 that are to be copied. : Linear elements L12-1, L12-2, L12-3: Divided linear elements L13-1, L13-2, L13-3: Divided linear elements L21, L22, L23, L31: Copy target Partial linear elements L51, L52: Linear elements L61 to L66: Linear elements or partial linear elements L71 to L76 to be copied: Linear elements or partial linear elements L811 to be copied L84: Linear element or partial linear element L91 to L95 to be copied: Linear element or partial linear element to be copied P0: Original image P0 (i, j): Pixel Pk constituting the original image : Correction work Image (k = 0,1,2,3, ...)
Pk (i, j): Pixels constituting the image for correction work Pn: Corrected images Q1 to Q9: Pixels and their center points S1 to S7: Steps in the flowchart Sk: Shift image (k = 1, 2, 3, …)
Sk (i, j): pixel SP constituting the shift image: gap area W: gap width Δ: shift amount

Claims (22)

An apparatus for correcting the density distribution of linear elements for an image formed by arranging a large number of linear elements,
An original image input unit for inputting an original image to be corrected as a binary image composed of an aggregate of black pixels constituting a linear element portion and white pixels constituting a background portion;
A correction work image storage unit for storing the input original image as an initial image;
A shift image storage unit for storing the input original image as an initial image;
For the image in the correction work image storage unit, a distance calculation unit for obtaining a distance to the nearest black pixel for each pixel;
A distance image storage unit for storing a distance image having a value indicating a distance calculated by the distance calculation unit as a pixel value;
A parameter setting unit for setting the predetermined shift amount Δ and the gap width W as parameters;
A shift processing unit that updates the image in the shift image storage unit by shifting the image by a shift amount Δ in a predetermined direction;
Among the black pixels of the image in the shift image storage unit, the black pixel whose pixel value at the corresponding position of the image in the distance image storage unit exceeds the gap width W is selected as the black pixel in the correction work image storage unit. A copy processing unit for copying to the corresponding position of the image;
A device for correcting an image having a linear element. The correction device according to claim 1,
The copy processing unit performs processing for obtaining the pixel value of the pixel at the corresponding position of the image in the distance image storage unit as each reference pixel value for each black pixel of the image in the shift image storage unit. When a black pixel that exceeds the gap width W is found, a process for replacing the white pixel of the image in the image storage unit for correction work at the position corresponding to the black pixel with a black pixel is performed. Image correction device having elements. The correction device according to claim 1,
For each white pixel of the image in the correction work image storage unit, the copy processing unit determines that “the pixel of the image in the shift image storage unit at the corresponding position is a black pixel” and “the pixel in the distance image storage unit at the corresponding position”. It is determined whether or not the condition that the pixel value of the pixel of the image exceeds the gap width W is satisfied, and when the condition is satisfied, a process of replacing the white pixel with a black pixel is performed. An image correction apparatus having a linear element. An apparatus for correcting the density distribution of linear elements for an image formed by arranging a large number of linear elements,
An original image input unit for inputting an original image to be corrected as a binary image composed of an aggregate of black pixels constituting a linear element portion and white pixels constituting a background portion;
A correction work image storage unit for storing the input original image as an initial image;
A shift image storage unit for storing the input original image as an initial image;
For the image in the correction work image storage unit, a distance calculation unit for obtaining a distance to the nearest black pixel for each pixel;
A parameter setting unit for setting the predetermined shift amount Δ and the gap width W as parameters;
A shift processing unit that updates the image in the shift image storage unit by shifting the image by a shift amount Δ in a predetermined direction;
Among the black pixels of the image in the shift image storage unit, the black pixel whose distance obtained by the distance calculation unit for the pixel at the corresponding position of the image in the correction work image storage unit exceeds the gap width W Is copied to the corresponding position of the image in the correction work image storage unit,
A device for correcting an image having a linear element. In the correction apparatus in any one of Claims 1-4,
The shift processing unit repeatedly performs the shift process only a plurality of times,
The copy processing unit executes a copy process (including an empty copy process where no substantial copy is performed at all) each time the shift process is performed,
An apparatus for correcting an image having linear elements, wherein the distance calculation unit executes a distance calculation process on the original image and a distance calculation process on the image after the copy process. The correction device according to claim 5,
Correction of an image having a linear element, wherein the shift processing section repeatedly executes the shift processing until the total number of pixels from which distance exceeding the gap width W is obtained by the distance calculation processing is equal to or less than a predetermined reference value apparatus. The correction device according to claim 5,
An apparatus for correcting an image having linear elements, wherein the shift processing section repeatedly executes the shift processing until the empty copy processing is continuously executed a predetermined number of times. In the correction apparatus in any one of Claims 1-7,
An apparatus for correcting an image having a linear element, wherein the distance calculation unit obtains, as a calculation value, a Euclidean distance between center points of two pixels or a value obtained by rounding the Euclidean distance by a predetermined significant digit. In the correction apparatus in any one of Claims 1-7,
An apparatus for correcting an image having a linear element, characterized in that a distance calculation unit obtains a 4-neighbor distance or an 8-neighbor distance between two pixels as a computed value. In the correction apparatus in any one of Claims 1-9,
An apparatus for correcting an image having a linear element, wherein the shift processing unit performs a rotation shift in which a pixel protruding from one side of the image by the shift is arranged in a blank portion generated in the other side of the image by the shift. In the correction apparatus in any one of Claims 1-10,
The parameter setting unit has a function of setting the shift direction as a parameter,
An image correcting apparatus having a linear element, wherein a shift processing unit executes a shift process in the shift direction.   The program for functioning a computer as a correction apparatus in any one of Claims 1-11. A method of correcting the density distribution of linear elements for an image formed by arranging a large number of linear elements,
The original image to be corrected is input to the computer as a binary image consisting of an aggregate of black pixels forming the linear element portion and white pixels forming the background portion, and the input original image is corrected in the initial state. Original image input stage to be set as an image for use and a shifted image in the initial state;
A parameter setting step for setting the predetermined shift amount Δ and the gap width W as parameters to the computer;
A distance calculating step for obtaining a distance from the nearest black pixel for each pixel of the correction work image in the initial state or the subsequent state;
An image shift step in which the computer shifts and updates the shifted image in the initial state or the subsequent state by the shift amount Δ in a predetermined direction;
The computer selects a black pixel in which the distance obtained in the distance calculation step for the pixel at the corresponding position of the correction work image among the black pixels of the shift image exceeds the gap width W. An image copy stage for copying and updating to the corresponding position of
A corrected image output stage in which the computer outputs the updated image for correction work as a corrected image;
A method for correcting an image having a linear element characterized by comprising: The correction method according to claim 13,
The corrected image output step is executed after the distance calculation step, the image shift step, and the image copy step (including an empty copy step in which no substantial copy is performed) are repeatedly executed a plurality of times. A method for correcting an image having a linear element. The correction method according to claim 14,
Correction of an image having a linear element, wherein a corrected image output step is executed when the total number of pixels that can obtain a distance exceeding the gap width W is equal to or less than a predetermined reference value in the distance calculation step Method. The correction method according to claim 14,
A method for correcting an image having linear elements, wherein the corrected image output step is executed when the empty copy step is continuously executed a predetermined number of times. In the correction method in any one of Claims 13-16,
A method for correcting an image having a linear element, characterized in that a Euclidean distance between center points of two pixels or a value obtained by rounding the Euclidean distance by a predetermined effective digit is obtained as a calculated value in a distance calculation stage. In the correction method in any one of Claims 13-16,
A method for correcting an image having a linear element, characterized in that, in a distance calculation stage, a four-neighbor distance or an eight-neighbor distance between two pixels is obtained as a calculated value. In the correction method in any one of Claims 13-18,
A method of correcting an image having a linear element, characterized in that, in the image shift stage, a rotation shift is performed in which a pixel protruding from one side of the image by the shift is arranged in a blank portion generated in the other side of the image by the shift. In the correction method in any one of Claims 13-19,
A method for correcting an image having linear elements, characterized in that, in the parameter setting stage, a value substantially equal to an average width value of each linear element is set as the gap width W. In the correction method in any one of Claims 13-20,
A method for correcting an image having linear elements, characterized in that a value twice as large as a gap width W is set as a shift amount Δ at a parameter setting stage. In the correction method in any one of Claims 13-21,
In the image shift stage, a rough longitudinal direction common to all linear elements is instructed to the computer, and the computer executes a shift process in which the direction orthogonal to the longitudinal direction is the shift direction. A method for correcting an image having a linear element.